O R I G I NA L A RT I C L E
doi:10.1111/j.1558-5646.2008.00511.x
GEOGRAPHIC VARIATION IN ADAPTATION AT
THE MOLECULAR LEVEL: A CASE STUDY OF
PLANT IMMUNITY GENES
David A. Moeller1,2 and Peter Tiffin3,4
1
Department of Genetics, University of Georgia, Athens, Georgia 30602
2
3
E-mail: [email protected]
Department of Plant Biology, University of Minnesota, Saint Paul, Minnesota 55108
4
E-mail: [email protected]
Received February 12, 2008
Accepted July 31, 2008
Natural selection imposed by interacting species frequently varies among geographic locations and can lead to local adaptation,
where alternative phenotypes are found in different populations. Little is known, however, about whether geographically variable
selection acting on traits that mediate species interactions is consistent or strong enough to influence patterns of nucleotide
variation at individual loci. To investigate this question, we examined patterns of nucleotide diversity and population structure at
16 plant innate immunity genes, with putative functions in defending plants against pathogens or herbivores, from six populations
of teosinte (Zea mays ssp. parviglumis). Specifically, we tested whether patterns of population structure and within-population
diversity at immunity genes differed from patterns found at nonimmunity (reference) loci and from neutral expectations derived
from coalescent simulations of structured populations. For the majority of genes, we detected no strong evidence of geographically
variable selection. However, in the wound-induced serine protease inhibitor (wip1), which inhibits the hydrolysis of dietary
proteins in insect herbivores, one population showed unusually high levels of genetic differentiation, very low levels of nucleotide
polymorphism, and was fixed for a novel replacement substitution in the active site of the protein. Taken together, these data
suggest that wip1 experienced a recent selective sweep in one geographic region; this pattern may reflect local adaptation or
an ongoing species-wide sweep. Overall, our results indicate that a signature of local adaptation at the molecular level may be
uncommon—particularly for traits that are under complex genetic control.
KEY WORDS:
Host–parasite interactions, local adaptation, natural selection, nucleotide variation, population structure.
Natural selection is a major force driving population divergence
in ecologically important traits (e.g., Endler 1977; Linhart and
Grant 1996; Mousseau et al. 2000; Reznick and Ghalambor
2001). Recently, the importance of spatially variable selection
and local adaptation has received considerable attention in studies
of species interactions and coevolution. This interest has been
motivated in part by the recognition that species interactions often
vary across landscapes (e.g., Kaltz and Shykoff 1998; Kraaijiveld
and Godfray 1999; Thompson and Cunningham 2002; Zangerl
and Berenbaum 2003; Rudgers and Strauss 2004; Heath and
C
3069
Tiffin 2007) and that traits that mediate species interactions are
often differentiated among populations in parallel with variation
in species interactions (e.g., Carrol and Boyd 1992; Brodie et al.
2002; Moeller 2006). Differences among populations in selection
imposed by interacting species may reflect geographic variation in
the abiotic and community context in which those interactions occur. Evidence of spatially variable selection and local adaptation
of traits involved in species interactions have been documented
at the phenotypic level (reviewed in Hoeksema and Forde 2008),
but there have been few attempts to determine whether the genes
C 2008 The Society for the Study of Evolution.
2008 The Author(s). Journal compilation Evolution 62-12: 3069–3081
D. A . M O E L L E R A N D P. T I F F I N
underlying these traits harbor evidence of geographically variable
selection.
At the molecular level, local adaptation occurs when alternative alleles (or multilocus genotypes) are favored in different
geographic locations, a process that would be expected to lead
to low levels of nucleotide and allelic variation within populations under directional selection (throughout this article we refer
to geographically distinct groups of intermating individuals as
“populations” although these are often referred to as “subpopulations” in the population genetics literature). By contrast, when
considering the broadest spatial scales (e.g., the species’ range),
geographical variation in selection and local adaptation has been
considered a form of balancing selection that maintains alternative
alleles within a species (Felsenstein 1976; Hedrick et al. 1976;
Hedrick 1986). Therefore, geographically variable selection may
lead to unusually low or high levels of nucleotide diversity depending on whether one examines diversity within a population or
combines samples from multiple populations inhabiting contrasting selective environments. Geographically variable selection can
also cause significantly higher levels of population genetic differentiation (e.g., F ST ) in focal (candidate) genes relative to other loci
in the genome (Lewontin and Krakauer 1973; Taylor et al. 1995;
Storz 2005; Stinchcombe and Hoekstra 2008)—a pattern that can
be tested only if diversity is sampled from multiple populations.
In principle, a signature of local adaptation at the molecular level
could be detectable from sequence data; however, recent theoretical work has indicated that nucleotide polymorphism may often
be unaffected by geographically variable selection or may be difficult to detect with standard tests of the neutral model of molecular
evolution, particularly when phenotypes are under complex genetic control (Kelly 2006).
In this article, we use a survey of nucleotide variation in 16
plant immunity genes from six natural populations of the annual
plant, teosinte (Zea mays ssp. parviglumis), to test for evidence of
geographic variation in the history of natural selection on genes
that play a role in defending plants against their natural enemies.
The plant innate immune system is comprised of a recognition
phase, which involves proteins that detect enemies, and a response phase, which involves proteins or secondary compounds
that inhibit attack by altering plant physiology or directly interacting with enemies or their targets in the host (Dangl and Jones
2001; Jones and Dangl 2006). Sequence variation in genes involved in the plant immune system has been well characterized in
several model systems (e.g., Arabidopsis thaliana, maize, tomato)
and has been increasingly studied in the wild relatives of these
model systems to test hypotheses about the process of coevolution
at the molecular level (reviewed in deMeaux and Mitchell-Olds
2003; Tiffin and Moeller 2006). The majority of these studies
have analyzed species-wide samples, and have revealed a variety
of evolutionary histories. Some genes, predominantly R-genes
3070
EVOLUTION DECEMBER 2008
involved in pathogen detection, show the signature of long-term
balanced polymorphism (e.g., Stahl et al. 1999; Tian et al. 2002;
Mauricio et al. 2003; Kroymann et al. 2003). Other immunity
genes show evidence of species-wide selective sweeps (Bishop
et al. 2000; Tiffin 2004; Tiffin et al. 2004), a pattern inconsistent
with local adaptation. The majority of genes involved in immune
response, however, harbor more complex patterns of sequence
variation that may reflect spatial and temporal heterogeneity in
the selective environment (Rose et al. 2004; Moeller and Tiffin
2005). In virtually all of these studies, the potential for geographically variable selection has been overlooked, or has been impossible to conclusively infer, because sequences were sampled from
across species’ ranges. Two exceptions, which have used surveys
of molecular diversity to look for evidence of local adaptation in
plant R genes, have not found strong evidence (de Meaux et al.
2003; Bakker et al. 2006).
The majority of the 16 genes we treat as immunity genes
have been previously identified to be upregulated in maize in response to leaf wounding or pathogen infection and/or have been
shown to function directly in resistance to pathogens or herbivores. These genes encode proteins with diverse biochemical
function including chitinases, protease inhibitors, and ribosomeinactivating proteins (see Table 1 for putative functions and references; also Moeller and Tiffin 2005). Fifteen of these genes code
for proteins that are classified as pathogenesis-related proteins;
their expression is induced following infection by pathogens or
damage by herbivores (Datta and Muthukrishnan 1999). Some
of the defense proteins that we studied function as enzymes to
confer resistance. For example, mir1 is a cysteine protease produced by some maize lines that confers resistance to a variety
of Lepidoptera by damaging the peritrophic matrix of caterpillar guts (Pechan et al. 2002). Other proteins function as enzyme
inhibitors. For example, mpi codes for a protease inhibitor in
maize that disrupts the function of digestive enzymes including
chymotrypsin in caterpillars (Spodoptera littoralis) and elastase
in mammals (Tamayo et al. 2000). Other proteins in our survey
are expressed in response to fungal pathogen infection. PRMs are
localized to the plasmodesmata of fungal-infected maize radicles
(Murillo et al. 1997) and chitinases are hydrolytic enzymes that
degrade chitin or glucan, major constituents of fungal cell walls
(e.g., Schlumbaum et al. 1986). Some of these genes may also
have roles in pathogen detection (e.g., chitinases). Previous analyses of these genes revealed unusually high levels of replacement
polymorphism relative to silent polymorphism in range-wide samples, a pattern that may result from selection favoring different
amino acids in different geographic locations, i.e. local adaptation
(Moeller and Tiffin 2005).
In this study, we examined sequence diversity within multiple geographically distinct populations to explicitly test for evidence of local adaptation of plant immunity genes. Geographical
G E O G R A P H I C VA R I AT I O N I N A DA P TAT I O N
Table 1. Summary statistics for range-wide samples of immunity and reference loci including the size of the sequenced fragment,
number of sequences (N), number of polymorphic sites (S), haplotype diversity (H d ), nucleotide diversity (π), the ratio of replacement
to silent polymorphism (π rep /π sil ), Tajima’s D, population genetic differentiation (F ST ), and the population recombination rate (Hudson’s
(1987) 4Nr scaled by π) across all populations, and the protein class and/or function if known.
Locus
Immunity
chiA
chiB
chiI
hag
hm2
mir1
mpi
plt2
pr1
pr5
pr6
prms
rip1
rip2
wip1
zlp
Reference
adh1
asg65
bnl7
bz2
fus6
glb
waxy
Totals
Hd
π
π rep /
π sil
D
109
74
46
55
63
71
25
37
22
27
41
38
103
41
45
27
0.993
0.989
0.988
0.958
0.889
0.980
0.948
0.963
0.971
0.963
0.991
0.796
0.973
0.988
0.956
0.942
0.0106
0.0130
0.0054
0.0094
0.0086
0.0123
0.0068
0.0069
0.0082
0.0102
0.0068
0.0091
0.0108
0.0091
0.0157
0.0038
0.373
0.311
0.088
0.441
0.367
0.452
0.516
0.444
0.111
1.018
0.308
0.376
0.436
0.394
0.208
0.139
−1.75
−0.55
−1.29
−1.50
−0.80
−1.06
−0.83
−1.52
−0.35
−0.26
−0.44
−0.68
−1.49
−1.12
−0.56
−1.54
0.142
12.1
0.188
4.9
0.155
48.1
0.270
3.0
0.046
0.6
0.145
4.1
0.196
17.5
0.120
12.3
0.183
63.2
0.194
4.3
0.132
22.8
0.157
0.0
0.207
5.7
0.255
62.3
0.316
15.7
0.183 1150.3
chitinase
chitinase
chitinase
thaumatin-like protein
HC-toxin degradation
cysteine protease
protease inhibitor
lipid transferase
function unknown
thaumatin-like protein
β-1,3 glucanase
pr1 class
ribosome-inactivating
ribosome-inactivating
protease inhibitor
thaumatin-like protein
EU724136–724216
EU274217–724289
EU724432–724507
EU724290–724364
EU724735–724809
EU724578–724661
EU724508–724577
EU724662–724734
EU724810–724890
EU725050–725133
EU724973–725049
EU724891–724972
EU725134–725209
EU725210–725284
EU725370–725455
EU725285–725369
1530
73 112
949
78
68
975
71
56
685
76
22
778
67
21
1266
81 184
1479
80
68
24961 1796 1355
0.987
0.972
0.949
0.896
0.924
0.991
0.978
0.0173
0.0097
0.0094
0.0076
0.0097
0.0214
0.0091
–
0.035
N/A
N/A
0.309
0.060
0.322
0.054
–
−0.21
−1.11
−0.83
0.21
−0.40
−1.57
−1.20
0.170
0.102
0.414
0.170
0.239
0.124
0.349
–
alcohol dehydrogenase
noncoding
noncoding
anthocyanin biosynthesis
RFLP marker
embryo storage protein
starch biosynthesis
–
EF539343–539415
EF539416–539493
EF539494–539564
EU724060–724135
EU724365–724431
EF539565–539645
EF539646–539725
Fragment N
size
1484
1273
1145
784
2536
1216
699
672
492
616
1510
673
1373
681
1382
763
S
81
73
76
75
76
84
70
73
81
84
77
83
76
75
87
85
variation in selection and local adaptation may be particularly
relevant in parviglumis because populations (1) occur across a
topographically variable region of southwest Mexico with strong
environmental gradients and (2) exhibit moderate-to-high levels
of neutral genetic differentiation and limited migration in some
regions (Moeller et al. 2007). If geographically variable selection
is more important in shaping diversity at the nucleotide level at
immunity than nonimmunity genes, we expect immunity loci to
have lower within-population diversity and higher population genetic structure than nonimmunity (reference) loci. We used DNA
sequence data from the 16 immunity genes along with seven
reference loci, which have no function in plant immunity, to test
whether (1) patterns of nucleotide diversity and population genetic
differentiation differ between the two classes of loci: immunity
vs. reference loci, (2) population genetic parameters for individual immunity loci differ significantly from distributions derived
from coalescent simulations of structured populations, and (3)
population genetic parameters for individual immunity loci differ
F ST
4Nr/π Protein Class/
Putative function
3.6
0.9
0.2
0.3
17.9
12.0
0.7
GenBank
accession #s
significantly from empirical distributions of population genetic
parameters observed in reference loci.
Materials and Methods
SAMPLING OF POPULATIONS AND LOCI
Zea mays ssp. parviglumis Iltis and Doebley is a highly outcrossing annual plant endemic to southwest Mexico. We examined sequence variation within and among six natural populations. Four
of these populations are in the western portion of its range (state of
Jalisco, Mexico) and two are in central portion of its range in the
Balsas River valley (state of Mexico, Mexico) (Fig. 1). These two
regions are separated by ∼300 km of mountainous terrain and the
populations representing the two regions correspond to the two
races of parviglumis distinguished by Wilkes (1967) based on
morphometric traits. Populations within each region are also geographically separated and occur across a wide range of elevations
and environments. Seeds were collected from 8 to 18 maternal
EVOLUTION DECEMBER 2008
3071
D. A . M O E L L E R A N D P. T I F F I N
Figure 1.
Map of Zea mays ssp. parviglumis study populations in
southwest Mexico; the inset map shows the current distribution
of the entire subspecies.
plants in each of the six populations. One seed per maternal plant
was germinated, grown to the seedling stage, and leaves were
harvested for DNA extraction using Dneasy plant kits (Qiagen,
Valencia, CA). We sampled 16 immunity loci and seven reference
(nonimmunity) loci that have no known role in plant resistance; all
loci were sampled from the same DNAs. Five of these reference
loci include protein-coding regions with diverse function in plant
growth and development and two loci are noncoding (Table 1).
An analysis of nucleotide polymorphism and population structure for five of these reference loci has been reported previously
(Moeller et al. 2007). Data from the remaining 18 loci have not
been reported elsewhere. All sequences have been submitted to
GenBank (Table 1).
CHARACTERIZATION OF SEQUENCE VARIATION
We used PCR to amplify the 23 immunity and reference loci from
each of 84 DNAs (Table 1). PCR products were either directly
sequenced or more often cloned into pGEM-T vectors (Promega,
Madison, WI) before sequencing. For each DNA sample, one
cloned fragment per locus was sequenced. Sequences were assembled and aligned in BioEdit 7.0.4.1 (Hall 1999). When singleton variants from cloned fragments were identified in alignments, they were checked by directly sequencing PCR products
or by sequencing multiple clones from multiple PCR reactions.
Sequenced regions ranged in size from 492 bp (pr1) to 2536
bp (hm2) (Table 1). For each population sample for each locus,
we used DNAsp 4.0 (Rozas et al. 2003) to calculated six commonly used estimators of sequence variation: haplotype diversity
(H d , calculated using eq. 8.4, in Nei 1987), nucleotide diversity (π, Nei 1987), Tajima’s D (Tajima 1989), population genetic
differentiation from other populations (F ST , based on genetic distance), population recombination rate (4Nr scaled by π) (Hudson
3072
EVOLUTION DECEMBER 2008
Figure 2.
Population-specific estimates of haplotype diversity,
nucleotide diversity (π), and population genetic differentiation
(pairwise F ST ) for each of the seven reference loci and the 16 immunity loci. Estimates for population T are highlighted by open
circles whereas the remaining populations are shown with filled
circles.
1987), and the relationship between within-species diversity and
between-species divergence (for which we used Tripsacum dactyloides (Poaceae) as an outgroup). For estimates of population
genetic differentiation, we found that values of F ST , a statistic
based on diversity within and between populations, were strongly
correlated with estimates of S nn , a statistic based on an allele’s
nearest-neighbor in a coalescent framework (Hudson 2000). Because the values of these two statistics were highly correlated and
produced qualitatively similar results, we present only estimates
of F ST . All descriptors of diversity and divergence were estimated
using DNAsp 4.0 (Rozas et al. 2003) and F ST estimates were made
using Arlequin (Schneider et al. 2000)
STATISTICAL ANALYSES
To test for geographic variation in the evolutionary history of
immunity genes, we took three analytical approaches. Our first
approach was to test for differences in nucleotide diversity and
population structure (H d , π, π rep / π sil , Tajima’s D, and F ST ) between the set of immunity and reference loci for both rangewide and population-specific samples. For range-wide samples,
we compared gene classes using nonparametric Wilcoxon rank
sum tests. For population-specific samples, we used the same
statistical tests but conducted separate analyses for each population. These analyses test whether polymorphism and population
G E O G R A P H I C VA R I AT I O N I N A DA P TAT I O N
structure at immunity loci deviate consistently from that of other
loci in the genome. If immunity loci are typically the subject
of geographically variable selection, then we expect significantly
lower sequence variation (H d and π), frequency spectra of polymorphism more strongly skewed toward rare variations (Tajima’s
D), and higher F ST at immunity relative to reference loci. If alternative alleles are favored by selection in different populations,
then we expect an elevated ratio of π rep / π sil in range-wide samples of immunity loci, but not in population-specific samples.
These tests for nonneutral evolution of immunity loci are conservative given that a significant difference between immunity
and reference loci is expected only if the effects of selection are
consistent across immunity loci. Moreover, if nonneutral evolution in the candidate (immunity) loci is not common, significant
differences will not likely be detected.
Our second approach to testing for local adaptation in immunity genes was to use a maximum-likelihood HKA test (mlHKA,
Wright and Charlesworth 2004), which examines the relationship
between within-species nucleotide polymorphism and betweenspecies sequence divergence for a set of candidate loci and reference loci (our analyses use sequence data from T. dactyloides
as an outgroup). The test uses maximum likelihood to determine whether models allowing the selection parameter, k, to vary
freely for immunity genes (but where reference loci are fixed to
evolve neutrally, k = 1) provide a significantly improved fit over
completely neutral models where k is fixed at 1 for all loci. We
conducted separate mlHKA tests for each population to evaluate
whether the evolutionary history of immunity genes has differed
among populations, consistent with geographically variable selection. From each analysis, we examined estimates of the selection
parameter, k, to determine whether polymorphism was elevated or
reduced relative to neutral expectations, suggestive of balancing
versus directional selection, respectively. For individual loci that
exhibited strong deviations from neutral expectations in the selection models (k > 2 or k < 0.5), we conducted separate analyses
allowing for k to vary freely only for the deviant locus.
Our third approach tested whether individual immunity loci
have extreme patterns of sequence variation and population structure in comparison to (1) distributions of population genetic
parameters based on coalescent simulations of subdivided populations and (2) empirical distributions of population genetic parameters based on the reference loci sampled in this study. As
described above, unusually high F ST and low sequence variation
within populations may indicate geographically variable selection. Values for immunity loci in the lower 5% of distributions for
nucleotide diversity and the upper 5% of the distributions for F ST
were considered candidates for local adaptation. Coalescent simulations of subdivided populations were conducted using Serial
SIMCOAL (Anderson et al. 2005). Serial SIMCOAL, like SIMCOAL (Excoffier et al. 2000) upon which it is based, provides a
framework for conducting coalescent simulations with multiple
populations connected by gene flow and undergoing population
growth or shrinkage. Population size and growth rate of each
population, and gene flow between populations can all vary. We
simulated six populations with initial population sizes of either
150,000 individuals (for two populations) or 250,000 individuals
(for four populations), 1000 bp loci, and per locus mutation rate of
μ = 10−6 . These estimates of population size, rates of population
growth, and rates of gene flow between pairs of populations were
parameterized according to the range of estimates obtained from
LAMARC (Kuhner 2006) using data from five of the reference
loci analyzed previously (Moeller et al. 2007). LAMARC assumes
that migration rates are at equilibrium and therefore that population structure has been constant. Because these assumptions are
likely violated in parviglumis and most other species, migration
rates may be biased (e.g., Moeller et al. 2007). However, the median and mode of the simulated distributions were very similar
to that of our empirical distributions for each of the parameters
examined (Figs. 3 and 4), suggesting that the simulations captured
patterns of diversity found in our populations. Because simulation
runs using different migration and population growth parameters
produced qualitatively similar results, we report the simulation
results from the run that provided the most conservative test for
geographically variable selection. We calculated pairwise F ST s
and nucleotide diversity for each population for each of the 1000
simulated datasets using Arlequin (Schneider et al. 2000).
Results
RANGE-WIDE PATTERNS OF POLYMORPHISM AND
POPULATION STRUCTURE
Our survey of 23 loci (∼ 25,000 bases) from an average of 78 individuals/locus identified 1355 polymorphic sites (Table 1). Our
range-wide samples harbored high haplotype diversity at both
immunity and reference loci (H d > 0.9 for 14 of the 16 immunity loci and five of six reference loci, Wilcoxon rank sum
test: Z = −0.201, P = 0.841; Table 1). Range-wide nucleotide
polymorphism (π) differed fourfold across immunity loci with
wip1 having the highest and zlp the lowest polymorphism (π =
0.0157 vs. 0.0038) and threefold among the reference loci; however, we found no significant difference in mean polymorphism
between classes (Z = 1.205, P = 0.228). The frequency spectrum of polymorphism, as revealed by Tajima’s D, also did not
differ significantly between classes (Z = 0.802, P = 0.423). As
with patterns of nucleotide diversity, the magnitude of population structure (global F ST across all populations) did not differ
significantly between immunity and reference loci (Z = 0.367,
P = 0.713). The estimates of population differentiation at individual immunity loci differed almost sevenfold with wip1 having
the highest and hm2 the lowest levels of F ST (0.316 vs. 0.046;
EVOLUTION DECEMBER 2008
3073
D. A . M O E L L E R A N D P. T I F F I N
Distributions of nucleotide diversity (π) and population genetic differentiation (pairwise F ST ) derived from coalescent simulations of subdivided populations (top panels) along with distributions of observed population-specific estimates of π and F ST for each
Figure 3.
of the immunity loci (bottom panels). Dashed lines indicate the lower 5% significance threshold for π and the upper 5% significance
threshold for F ST . Arrows show the values of each population for wip1; values for population T are highlighted using open arrows. The
identity of outlier immunity loci is described in the Results.
Distributions of nucleotide diversity (π) and population genetic differentiation (pairwise F ST ) from population-specific estimates for each of the reference loci (top panels) along with distributions of observed population-specific estimates of π and F ST for each
of the immunity loci (bottom panels). Dashed lines indicate the lower 5% significance threshold for π and the upper 5% significance
Figure 4.
threshold for F ST . Arrows show the values of each population for wip1; values for population T are highlighted using open arrows. The
identity of outlier immunity loci is described in the Results.
3074
EVOLUTION DECEMBER 2008
G E O G R A P H I C VA R I AT I O N I N A DA P TAT I O N
Table 1); this variance was greater than the fourfold range of F ST
values at reference loci. The one significant difference between
immunity and reference loci that we detected was in the ratio of
replacement to silent polymorphism, which was higher in immunity than reference loci (Z = −2.35, P = 0.018). This excess
of replacement polymorphisms is consistent with findings from
a smaller range-wide sample of parviglumis (Moeller and Tiffin
2005) and may be due to different alleles having been selected in
different populations, that is geographically variable selection.
POPULATION-SPECIFIC PATTERNS OF
POLYMORPHISM AND POPULATION STRUCTURE
Population-specific estimates of haplotype and nucleotide diversity tended to be lower than range-wide estimates, but comparisons between gene classes yielded similar patterns to those found
in the range-wide samples. As with range-wide estimates, we did
not find significant differences between immunity and reference
loci for haplotype diversity (Wilcoxon rank sum test: P > 0.61
for all tests), nucleotide diversity (P > 0.08 for all tests), Tajima’s
D (P > 0.07 for all tests), or pairwise F ST (P > 0.36 for all tests)
within any of the six populations (Fig. 2). The ratio of replacement to silent polymorphism was significantly higher in immunity
than reference loci in two of the six populations, G and SNT (P
= 0.01 and P = 0.05, respectively), a pattern also found in the
range-wide sample. In summary, our comparison of immunity to
reference loci revealed no indication that immunity genes, as a
class, are frequently targets of geographically variable selection.
Instead, we found wide variation in the evolutionary history of
loci in both classes.
MAXIMUM-LIKELIHOOD HKA TESTS WITHIN
POPULATIONS
Maximum-likelihood HKA tests revealed no evidence that a
model allowing for selection on all immunity genes provided
a significantly improved fit over a completely neutral model for
any population or across all populations combined (Table 2). Testing all immunity loci together, however, may mask evidence of
selection that has acted on individual loci. Separate mlHKA tests
for each of the loci that deviated strongly from neutral expectations in the initial analyses (k > 2 or k < 0.05, Table 2) identified
three loci that deviated significantly from expected patterns under neutrality: polymorphism was strongly reduced for wip1 in
population T (χ2 = 5.52, df = 1, P < 0.01), and polymorphism
was significantly elevated for hag in population SNT (χ2 = 3.97,
df = 1, P < 0.05), and for pr1 in population G (χ2 = 3.48, df =
1, P < 0.05). The remaining tests for individual loci (hag in T,
pr1 in S, SNT, and T; mpi in T, chiA in S, and chiB in T) did
not indicate significant differences between neutral and selection
models. As with comparisons between immunity and reference
loci for diversity and F ST , this analysis showed that there was
considerable variation in the evolutionary history of the genes
examined, with the majority of genes having patterns of diversity
Maximum-likelihood HKA tests including seven immunity and seven reference loci for each of the six parviglumis populations
and the entire range-wide sample. In the upper panel, log-likelihood values are shown from a completely neutral model and from a
Table 2.
model allowing for the selection parameter, k, to vary freely for immunity loci; we did not find a significantly better fit for selection
models compared to neutral models for any population. The lower panel shows the selection parameter, k, for each immunity gene from
selection models. When values of k exceeded 2 or were less than 0.5 (shown in bold), an additional mlHKA test was run allowing for
selection only on the locus deviating from neutral expectations. Asterisks denote the cases in which a neutral model could be rejected
with ∗ P<0.05 or ∗∗ P<0.01, and for these loci, the log likelihoods are also shown.
Populations
A
Neutral model (ln L)
Selection model (ln L)
−80.09
−80.66
Immunity Genes
Species-wide
G
−80.76
−78.96
S
−77.85
−77.85
SNT
−83.50
−81.41
T
Z
−74.84
−70.04
−81.84
−82.00
−92.13
−93.10
2.02
1.38
1.49
4.58
1.75
2.66
2.59
1.39
2.44
1.93
0.00∗∗
(−72.08)
0.87
0.83
1.23
1.14
1.18
0.92
1.26
0.76
1.27
1.11
k values
hag
0.85
1.86
1.79
pr1
1.82
3.81
mpi
chiA
chiB
rip1
wip1
0.59
0.81
0.83
1.72
1.80
2.91∗
(−79.02)
0.93
0.83
1.16
1.74
1.12
4.78∗
(−80.66)
3.13
0.70
2.01
1.48
0.96
1.18
0.54
0.94
1.17
1.34
0.52
EVOLUTION DECEMBER 2008
3075
D. A . M O E L L E R A N D P. T I F F I N
indistinguishable from the reference loci, but with three immunity genes deviating from neutral expectations in one of the
populations.
TESTS USING COALESCENT AND EMPIRICAL
These patterns of nucleotide variation resulted in unusually high
estimates of genetic differentiation (F ST ) at wip1 between population T and other populations; whereas genetic differentiation
for wip1 between other pairs of populations (excluding T) were
not unusually high or approaching significance.
DISTRIBUTIONS
To further test whether individual immunity loci have been
the targets of geographically variable selection, we compared
population-specific estimates of π and pairwise F ST to distributions based on (1) coalescent simulations of structured, growing
populations (hereafter “coalescent distribution”) and (2) the collection of population-specific estimates of parameters from the
reference loci (hereafter “empirical distribution”). For nucleotide
diversity (π), two immunity loci were in the lower tail of the coalescent distribution as well as the empirical distribution (wip1,
popn T P = 0.0002; zlp, popn SNT P = 0.00155). One other
value fell into the lower 5% of the coalescent but not the empirical distribution (pr5 popn T, π = 0.0022). Of these three loci, only
wip1 in population T was also identified as having unusually low
nucleotide polymorphism in the mlHKA analyses (see above).
Values of F ST for seven pairwise comparisons between populations (involving three immunity genes) were in the upper
tail of both the coalescent and empirical distributions, indicating unusually high population structure for these genes. Although this frequency of extreme values is approximately what
would be expected by chance, four of these values were for
wip1 and involved comparisons between population T and other
populations (F ST = 0.53 – 0.76). A fifth F ST value for wip1
involving population T (and G) was in the upper 6% of both
the coalescent and empirical distributions (wip1, G vs. T, F ST =
0.48). Of the remaining extreme F ST values, two were for pr5
and involved comparisons between population T and two other
populations (G and S, F ST = 0.59 and 0.51). The remaining significant value was for hag and involved a comparison between
populations A and S (F ST = 0.58). The F ST for pr5 between populations S and T was in the upper tail of the coalescent but not
the empirical distribution. Comparisons between immunity loci
and either coalescent or empirical distributions were remarkably
similar, with the empirical distributions identifying slightly fewer
extreme values for immunity loci.
Across all tests, only wip1 showed strong evidence for geographically variable selection. For wip1, we found very low
nucleotide diversity and only two haplotypes in population T
(with one haplotype represented by a single sequence and differing from the other at a single silent site in the flanking region);
whereas, nucleotide diversity (π) was between 51- and 112-fold
higher in the remaining five populations (see Fig. 5). Most notably wip1 alleles in population T differed from nearly all other
wip1 alleles by a 565-bp insertion in the 3 flanking region and
a replacement substitution in the active site of the wip1 protein.
3076
EVOLUTION DECEMBER 2008
Discussion
GEOGRAPHIC VARIATION IN SELECTION AT THE
MOLECULAR LEVEL
Species interactions and patterns of selection on ecologically important traits frequently vary among geographic locations. The
extent to which local selection shapes nucleotide diversity at the
genes that mediate these interactions, however, is not well understood. If spatial variation in selection has shaped patterns of variation at the molecular level, then we expect population structure
and nucleotide diversity will differ significantly between genes
subject to local selection and those that are not subject to local
selection. Alternatively, if the spatial pattern and nature of selection have been weak or unstable through time, then geographic
variation in selection may leave no clear signature. In this study
we searched for evidence of local adaptation at genes involved in
protecting plants against pathogens and herbivores. This problem
is of particular interest for defense genes because evolutionary
ecologists have documented that geographically distinct populations often harbor different defense phenotypes and experience
different selection in contemporary populations.
Of the 16 immunity genes sampled from six parviglumis
populations, we detected evidence for geographically variable selection for only one immunity gene, the wound-inducible serine
protease inhibitor (wip1). Like many other protease inhibitors,
wip1 confers plant resistance to insect herbivores by inhibiting
chymotrypsin proteases in insect guts, and thus interferes with
the hydrolysis of dietary proteins (Rohrmeier and Lehle 1993).
Patterns of nucleotide variation within populations and divergence between populations both suggest a historical pattern of
geographically variable selection on wip1. First, range-wide nucleotide polymorphism at wip1 was the highest among all immunity loci, and five of the six populations harbored high levels of
variation similar to that observed in the range-wide sample. By
contrast, polymorphism at wip1 in population T was over 50-fold
lower than the remaining populations, had the lowest nucleotide
diversity of all population samples of immunity genes, and was
an outlier in both coalescent and empirical distributions based
on reference loci. Maximum-likelihood HKA analyses confirmed
that wip1 had significantly reduced polymorphism in population
T relative to other immunity and reference loci sampled in this
same population and lower levels of polymorphism than expected
in the absence of selection. These results suggest the action of directional selection on wip1 within population T. Second, genetic
G E O G R A P H I C VA R I AT I O N I N A DA P TAT I O N
Figure 5.
Gene structure and polymorphism in wip1 for each of the six populations. The length of the aligned sequences is 1382 bp
with one intron and 45 segregating sites. The two indels segregating in our population sample are shown with gray bars and the only
replacement substitution in the active site of the protein is highlighted by black boxes.
differentiation (F ST ) at wip1 between population T and other populations included some of the highest values measured and was
significantly higher than predicted on the basis of coalescent simulations and the empirical distribution based on reference loci.
Genetic differentiation at wip1 between other pairs of populations (excluding T) was not unusually high or approaching significance. These results suggest that population structure at wip1
differs strongly from that observed for other loci, consistent with
geographically variable selection.
A closer examination of sequence variation in wip1 showed
that population T harbored a single segregating site across all
1205 bases; the remaining populations had 11–35 segregating
sites (Fig. 5). Most notably, when compared to the remaining
populations, all individuals in population T differed from nearly
all other sequences by a 565 bp insertion in the 3 flanking region of the gene and at a replacement substitution resulting in a
nonpolar to polar amino acid change (glycine to serine) immediately adjacent to the active site of the protein (Fig. 4). The close
proximity of this amino-acid change to the molecule’s active site
may alter the stereochemical fit between the wip1 protein and chymotrypsin proteases, which in turn is expected to alter the efficacy
of enzyme inhibition (Rohrmeier and Lehle 1993). Analyses of
wip1 divergence among grass lineages has also revealed evidence
of adaptive evolution in the region of the protein that is near the
active site (Tiffin and Gaut 2001). Interestingly, the population
T allele is also found at low frequency in population G (Fig. 5).
A previous analysis of five nuclear loci not involved in host defense revealed evidence for directional migration primarily from
T to G (Moeller et al. 2007), suggesting that gene flow may have
introduced this novel allele into population G.
Patterns of diversity at wip1 suggest one of two possible evolutionary scenarios. One possibility is that these patterns reflect
EVOLUTION DECEMBER 2008
3077
D. A . M O E L L E R A N D P. T I F F I N
local adaptation, either because natural selection favors a different allele in population T than in other populations or that wip1 is
under positive selection in population T but not in the other populations. Because population T is found at a lower elevation than
the other populations we surveyed, historical biotic environments
in this population may have differed from other sampled populations. Under this scenario, the population T allele may be found
in population G because of recent migration but may be neutral
or deleterious in that environment. A second possible explanation
for the patterns of diversity at wip1 is that the novel wip1 allele
found in population T confers an advantage across all populations
but has yet to spread to fixation due to limited gene flow, particularly among Jalisco populations (Moeller et al. 2007). Under this
scenario, the recent selective sweep in population T does not reflect geographically variable selection but rather the initial phase
of an ongoing selective sweep. In structured populations, the time
it takes for a favorable allele to spread to fixation will depend not
only on the initial frequency and selective advantage, but also the
time it takes for migration to move the allele across populations
(Cherry and Wakeley 2003).
These two scenarios could be differentiated using field experiments if current biotic environments reflect the historical selective environments that drove differentiation. It is not clear,
however, that this is a reasonable assumption. Herbivore and disease pressure are highly variable in nature (e.g., Root 1996) and
therefore selection on plant defenses may be inconsistent through
time. Moreover, the acquisition of novel plant defenses through
adaptive evolution may reduce or eliminate enemy pressure in
natural populations, leading to an important shift in the selective environment. Therefore, if the enemy pressure that caused
selection on wip1 has abated, then field studies in current environments may not represent the environment in which adaptation
occurred. Both of these issues pose challenges for making connections between ecological process and evolutionary change at
the molecular level.
Patterns of sequence variation in two other genes may be
consistent with geographically variable selection, but the results
are much weaker than those for wip1. We found significantly high
levels of F ST between population A and some of the remaining
populations at hag, which encodes a thaumatin-like protein, along
with evidence that the frequency spectrum of polymorphism harbored an excess of derived mutations in population A but no
others (Fay and Wu’s H = −7.85, P < 0.05). Similarly, for pr5,
we found stronger than expected genetic differentiation between
population T and a subset of other populations, along with nucleotide polymorphism that was significantly lower than expected
(π = 0.0022, P < 0.05) and 4- to 5-fold less than that of the other
populations. Regardless of whether patterns of sequence variation
at these two genes are viewed as indicative of local adaptation, it
is clear that we found little convincing support for geographically
3078
EVOLUTION DECEMBER 2008
variable selection in the majority of immunity genes. In one case,
patterns of sequence variation appear to conform to predictions
under a range-wide selective sweep. For hm2 we found unusually
low levels of genetic divergence among populations, consistent
with previous results indicating that hm2 has been a target of positive selection in parviglumis (Zhang et al. 2002). Overall, these
results suggest that species interactions in this system may not
often be characterized by strong and consistent spatially variable
patterns of selection, or that their imprint on the genome is too
weak to detect by available methods.
EVALUATING EVIDENCE FOR GEOGRAPHICALLY
VARIABLE SELECTION
To date, most evidence for geographically variable selection at
the molecular level has come from studies of animals, particularly humans. For example, human populations have unusually
high divergence (F ST values) at genes influencing disease resistance (Tishkoff et al. 2001; Hamblin et al. 2002), lactose intolerance (Hollox et al. 2001), skin pigmentation (Rana et al.
1999), diabetes susceptibility (Fullerton et al. 2002), and behavior (Gilad et al. 2002). Most of these studies identified genes that
were a priori predicted to show adaptive population differentiation due to observed differences in phenotypes; although more
recent analyses have sought to use genome scans to identify a
suite of loci that were not a priori suspected to be the targets
of geographically variable selection (e.g., Akey et al. 2007). In
some cases, extreme levels of population structure (F ST ) have
been identified by comparing observed values to distributions of
population genetic parameters derived from coalescent simulation
(e.g., Bowcock et al. 1991). This model-based test for local selection has been criticized because simulation results are sensitive
to variation in population demographic history. Incorporating information on migration structure and changes in population size,
as we have done here, allow for more realistic simulation-based
tests, compared to models that assume stable population size and
panmixis (e.g., Storz et al. 2004). In our simulations, the median
and mode of the coalescent-based distributions of nucleotide diversity and population differentiation was very similar to empirical distributions, suggesting that our model-based tests for local
selection are unlikely to have been biased.
Indirect model-based tests for geographically variable selection are increasingly being replaced by the use of empirical distributions of population genetic parameters, particularly in model
organisms where genome-wide polymorphism data are now available. For example, Akey et al. (2007) used an empirical distribution of F ST based on SNP variation in humans to identify 156
candidate genes that may have been shaped by geographically
variable selection. The application of this direct analytical approach remains challenging for nonmodel organisms, where highdensity genome-wide polymorphism data are not yet available for
G E O G R A P H I C VA R I AT I O N I N A DA P TAT I O N
population samples. Although our study included relatively few
reference genes, our tests involving empirical distributions based
on our reference loci produced very similar results to tests involving coalescent distributions. The few studies to date that have detected nucleotide-level signatures of local adaptation in nonmodel
organisms have compared spatial patterns of variation in candidate loci to spatial patterns in sets of unlinked loci (e.g., Hoekstra
et al. 2004; Storz and Dubach 2004; Hemmer-Hansen et al. 2007;
Storz et al. 2007) or examined genome-wide variation in anonymous markers such as AFLPs rather than sequence variation in
candidate genes of known function (e.g., Emelianov et al. 2004;
Egan et al. 2008). Although these approaches to testing for local
adaptation at the molecular level have rarely been taken in studies
of plants, one study identified candidate genes underlying adaptation to drought and salt tolerance in Helianthus annuus (Kane
and Rieseberg 2007). A second study by deMeaux et al. (2003)
did not find differences in population structure between plant resistance genes and RAPD markers, but found some inconsistency
between patterns of population structure at the molecular level
and in resistance phenotypes. As genome-wide polymorphism
datasets become available for population samples, our ability to
distinguish geographically variable selection from other phenomena such as demographic history will be greatly improved.
LIMITS TO DETECTING GEOGRAPHICALLY VARIABLE
SELECTION
Kelly (2006) has recently shown that both biological and statistical factors may limit the ability to detect selection on genes
influencing quantitative traits under geographically variable selection. Using simulations of sequence evolution in the flanking
regions of QTL that underlie complex traits, Kelly (2006) showed
that selection that favored different phenotypes in different populations strongly affected quantitative genetic variation but rarely
affected nucleotide diversity in a manner that was detectable using
standard tests of nonneutral evolution. Based on these results, it is
likely that surveys of nucleotide diversity at genes that contribute
to variation in polygenic traits, such as the one presented here,
may underestimate the frequency with which phenotypes are the
subjects of local selection. These results are also important because they suggest that local adaptation may often occur through
standing variation, rather than being dependent on new mutations
or alleles that enter populations through migration.
The immunity loci in our study are all known to be upregulated in response to and/or effective in defending against natural
enemies, but the complexity of the genetics underlying phenotypes is not known for most of them. Even in cases in which
a biochemical function can be assigned to a protein, we do not
know how these functions are integrated into a phenotype upon
which selection acts. It is likely that many of the genes included
in our study (with the possible exception of hm2) contribute to a
multilocus phenotype. In addition, these genes may be pleiotropic
and confer resistance to multiple enemies, further complicating
the likelihood of detecting local selection in gene sequences. This
genetic architecture may have limited our ability to detect signatures of geographically variable selection at the molecular level,
even if selection has been stable over many generations.
ACKNOWLEDGMENTS
We thank Nicholas Lauter and Andy Muncaski for their help collecting
parviglumis seeds, N. Lauter and Jesus Sanchez for generously providing
us with seeds, and Maurine Neiman and two anonymous reviewers for
comments that improved the presentation of the work. Financial support
was provided by a grant from the National Science Foundation (DEB
0235027 to PT).
LITERATURE CITED
Akey, J. M., G. Zhang, K. Zhang, L. Jin, and M. D. Shriver. 2007. Interrogating
a high-density SNP map for signatures of natural selection. Genome Res.
12:1805–1814.
Anderson, C. N. K., U. Ramakrishnan, Y. L. Chan, and E. A. Hadly. 2005.
Serial SimCoal: a population genetic model for data from multiple populations and points in time. Bioinformatics 21:1733–1734.
Bakker, E. G., C. Toomajian, M. Kreitman, and J. Bergelson. 2006. Genomewide survey of R gene polymorphisms in Arabidopsis. Plant Cell
18:1803–1818.
Bishop, J. G., A. M. Dean, and T. Mitchell-Olds. 2000. Rapid evolution in plant
chitinases: molecular targets of selection in plant-pathogen coevolution.
Proc. Natl. Acad. Sci. USA 97:5322–5327.
Bowcock, A. M., J. R. Kidd, J. L. Mountain, J. M. Hebert, L. Carotenuto, K. K.
Kidd, and L. L. Cavalli-Sforza. 1991. Drift, admixture, and selection in
human evolution: a study with DNA polymorphisms. Proc. Natl. Acad.
Sci. USA 88:839–843.
Brodie, E. D., Jr., B. J. Ridenhour, and E. D. Brodie III. 2002. The evolutionary
response of predators to dangerous prey: hotspots and coldspots in the
geographic mosaic of coevolution between garter snakes and newts.
Evolution 56:2067–2082.
Carroll, S. P., and C. Boyd. 1992. Host race radiation in the soapberry bug:
natural history with the history. Evolution 46:1052–1069.
Cherry, J. L., and J. Wakeley. 2003. A diffusion approximation for selection
and drift in a subdivided population. Genetics 163:421–428.
Dangl, J. L., and J. D. G. Jones. 2001. Plant pathogens and integrated defence
responses to infection. Nature 411:826–833.
Datta, S. K., and S. Muthukrishnan. 1999. Pathogenesis-related proteins in
plants. CRC Press, Boca Raton, FL.
de Meaux, J., and T. Mitchell-Olds. 2003. Evolution of plant resistance at
the molecular level: ecological context of species interactions. Heredity
91:345–352.
de Meaux, J., I. Cattan-Toupance, C. Lavigne, T. Langin, and C. Neema. 2003.
Polymorphism of a complex resistance gene candidate family in wild
populations of common bean (Phaseolus vulgaris) in Argentina: comparison with phenotypic resistance polymorphism. Mol. Ecol. 12:263–273.
Egan, S.P., P. Nosil, and D.J. Funk. 2008. Selection and genomic differentiation during ecological speciation: isolating the contributions of host
association via a comparative genomic scan of Neochlamisus bebbianae
leaf beetles. Evolution 62:1162–1181.
Emelianov, I., F. Marec, and J. Mallet. 2004. Genomic evidence for divergence
with gene flow in host races of the larch budmoth. Proc. R. Soc. Lond.
B 271:97–105.
EVOLUTION DECEMBER 2008
3079
D. A . M O E L L E R A N D P. T I F F I N
Endler, J. A. 1977. Geographic variation, speciation, and clines. Princeton
Univ. Press, Princeton, NJ.
Excoffier, L., J. Novembre, and S. Schneider. 2000. SIMCOAL: a general
coalescent program for simulation of molecular data in interconnected
populations with arbitrary demography. J. Heredity 91:506–509.
Felsenstein, J. 1976. Theoretical population genetics of variable selection and
migration. Annu. Rev. Genet. 10:253–280.
Fullerton, S. M., A. Bartoszewicz, G. Ybazeta, Y. Horikawa, G. I. Bell,
K. K. Kidd, N. J. Cox, R. R. Hudson, and A. Di Rienzo. 2002. Geographic and haplotype structure of candidate type 2 diabetes susceptibility variants at hte calpain-10 locus. Am. J. Hum. Genet. 70:1096–
1106.
Gilad, Y., S. Rosenberg, M. Przeworski, D. Lancet, and K. Skorecki. 2002.
Evidence for positive selection and population structure at the human
MAO-A gene. Proc. Natl. Acad. Sci. USA 99:862–867.
Hall, T. A. 1999. BioEdit: a user-friendly biological sequence alignment editor
and program for windows 95/98/NT. Nucleic Acids Symp. Ser. 41:95–
98.
Hamblin, M. T., E. E. Thompson, and A. Di Rienzo. 2002. Complex signatures
of natural selection at the Duffy blood group locus. Am. J. Hum. Genet.
70:369–383.
Heath, K. D., and P. Tiffin. 2007. Context dependence in the coevolution of
plant and rhizobial mutualists. Proc. R. Soc. London B 274:1905–1912.
Hedrick, P. W. 1986. Genetic polymorphism in heterogeneous environments:
a decade later. Annu. Rev. Ecol. Syst. 17:535–566.
Hedrick, P. W., M. E. Ginevan, and E. P. Ewing. 1976. Genetic polymorphism
in heterogeneous environments. Annu. Rev. Ecol. Syst. 7:1–32.
Hemmer-Hansen, J., E. E. Nielsen, J. Frydenberg, and V. Loeschcke. 2007.
Adaptive divergence in a high gene flow environment: Hsc70 variation in the European flounder (Platichthys flesus L.). Heredity 99:592–
600.
Hoeksema, J. D., and S. E. Forde. 2008. A meta-analysis of factors affecting
local adaptation between interacting species. Am. Nat. 171:275–290.
Hoekstra, H. E., K. E. Drumm, and M. W. Nachman. 2004. Ecological genetics
of adaptive color polymorphism in pocket mice: geographic variation in
selected and neutral genes. Evolution 58:1329–1341.
Hollox, E. J., M. Poulter, M. Zvarik, V. Ferak, A. Krause, T. Jenkins, N. Saha,
et al. 2001. Lactase haplotype diversity in the Old World. Am. J. Hum.
Genet. 68:160–172.
Hudson, R. R. 1987. Estimating the recombination parameter of a finite population model without selection. Genet. Res. 50:245–250.
———. 2000. A new statistic for detecting genetic differentiation. Genetics
155:2011–2014.
Jones, J. D. G., and J. L. Dangl. 2006. The plant immune system. Nature
444:323–329.
Kaltz, O., and J. A. Shykoff. 1998. Local adaptation in host-parasite systems.
Heredity 81:361–370.
Kane N. C., and L. H. Rieseberg. 2007. Selective sweeps reveal candidate
genes for adaptation to drought and salt tolerance in common sunflower,
Helianthus annuus. Genetics 175:1823–1834
Kelly, J. K. 2006. Geographical variation in selection, from phenotypes to
molecules. Am. Nat. 167:481–495.
Kraaijeveld, A. R., and H. C. J. Godfray. 1999. Geographic patterns in the
evolution of resistance and virulence in Drosophila and its parasitoids.
Am. Nat. 153:S61–S74.
Kroymann, J., S. Donnerhacke, D. Schnabelrauch, and T. Mitchell-Olds. 2003.
Evolutionary dynamics of an Arabidopsis insect resistance quantitative
trait locus. Proc. Natl. Acad. Sci. USA 100:14587–14592.
Kuhner, M. K. 2006. LAMARC 2.0: maximum likelihood and Bayesian estimation of population parameters. Bioinformatics 22:768–770.
Lewontin, R. C., and J. Krakauer. 1973. Distribution of gene frequency as a
3080
EVOLUTION DECEMBER 2008
test of the theory of the selective neutrality of polymorphisms. Genetics
74:175–195.
Linhart, Y. B., and M. C. Grant. 1996. Evolutionary significance of local
genetic differentiation in plants. Annu. Rev. Ecol. Syst. 27:237–277.
Mauricio, R., E. A. Stahl, T. Korves, D. Tian, M. Kreitman, and J. Bergelson.
2003. Natural selection for polymorphism in the disease resistance gene
Rps2 of Arabidopsis. Genetics 163:735–746.
Moeller, D. A. 2006. Geographic structure of pollinator communities, reproductive assurance, and the evolution of self-pollination. Ecology
87:1510–1522.
Moeller, D. A., and P. Tiffin. 2005. Genetic diversity and the evolutionary
history of plant immunity genes in two species of Zea. Mol. Biol. Evol.
22:2480–2490.
Moeller, D. A., M. I. Tenaillon, and P. Tiffin. 2007. Population structure and
its effects on patterns of nucleotide polymorphism in teosinte (Zea mays
ssp. parviglumis). Genetics 176:1799–1809.
Mousseau, T. A., B. Sinervo, and J. A. Endler. 2000. Adaptive genetic variation
in the wild, Oxford Univ. Press, Oxford, U.K.
Murillo, I., L. Cavallarin, and B. San Segundo. 1997. The maize pathogenesisrelated PRms protein localizes to plasmodesmata in maize radicles. Plant
Cell 9:145–156.
Nei, M. 1987. Molecular evolutionary genetics. New York, Columbia Univ.
Press, New York.
Pechan, T., A. Cohen, W. P. Williams, and D. S. Luthe. 2002. Insect feeding
mobilizes a unique plant defense protease that disrupts the peritrophic
matrix of caterpillars. Proc. Nat. Acad. Sci., USA 99:13319–13323.
Rana, B. K., D. Hewett-Emmett, L. Jin, B. H. Chang, N. Sammbuughin,
M. Lin, S. Watkins, et al. 1999. High polymorphism at the human
melanocortin 1 receptor locus. Genetics 151:1547–1557.
Reznick, D. N., and C. K. Ghalambor. 2001. The population ecology of
contemporary adaptations: what empirical studies reveal about the conditions that promote adaptive evolution. Genetica 112:183–198.
Rohrmeier, T., and L. Lehle. 1993. WIP1, a wound-inducible gene from maize
with homology to Bowman-Birk proteinase inhibitors. Plant Mol. Biol.
22:783.
Root, R. B. 1996. Herbivore pressure on goldenrods (Solidago altissima): its
variation and cumulative effects. Ecology 77:1074–1087.
Rose, L. E., P. D. Bittner-Eddy, C. H. Langley, E. B. Holub, R. W. Michelmore,
and J. L. Beynon. 2004. The maintenance of extreme amino acid diversity
at the disease resistance gene, RPP13, in Arabidopsis thaliana. Genetics
166:1517–1527.
Rozas, J., J. C. Sánchez-DelBarrio, X. Messeguer, and R. Rozas. 2003. DnaSP,
DNA polymorphism analyses by the coalescent and other methods.
Bioinformatics 19:2496–2497.
Rudgers, J. A., and S. Y. Strauss. 2004. A selection mosaic in the facultative mutualism between ants and wild cotton. Proc. R. Soc. Lond. B
271:2481–2488.
Schneider, S., D. Roessli, and L. Excoffier. 2000. ARLEQUIN: a software for
population genetics data analysis, version 2.0.Genetics and Biometry
Laboratory, Univ. of Geneva.
Stahl, E. A., G. Dwyer, R. Mauricio, M. Kreitman, and J. Bergelson. 1999.
Dynamics of disease resistance polymorphism at the Rpm1 locus of
Arabidopsis. Nature 400:667–671.
Stinchcombe, J.R., and H.E. Hoekstra. 2008. Combining population genomics
and quantitative genetics: finding the genes underlying ecologically important traits. Heredity 100:158–170.
Storz, J. F. 2005. Using genome scans of DNA polymorphism to infer adaptive
population divergence. Mol. Ecol. 14:671–688.
Storz, J. F., and J. M. Dubach. 2004. Natural selection drives altitudinal
divergence at the albumin locus in deer mice, Peromyscus maniculatus.
Evolution 14:671–688.
G E O G R A P H I C VA R I AT I O N I N A DA P TAT I O N
Storz, J. F., B. A. Payseur, and M. W. Nachman. 2004. Genome scans of DNA
variability in humans reveal evidence for selective sweeps outside of
Africa. Mol. Biol. Evol. 21:1800–1811.
Storz, J.F., S.J. Sabatino, F.G. Hoffmann, E.J. Gering, H. Moriyama, et al.
2007. The molecular basis of high-altitude adaptation in deer mice.
PLoS Genet. 3: e45, 0448–0459.
Tajima, F. 1989. Statistical method for testing the neutral mutation hypothesis
by DNA polymorphism. Genetics 123:585–595.
Tamayo, M. C., M. Rufat, J. M. Bravo, and B. San Segunto. 2000. Accumulation of a maize proteinase inhibitor in response to wounding and insect
feeding, and characterization of its activity toward digestive proteinases
of Spodoptera littoralis larvae Planta 211:62–71.
Taylor, M. F. J., Y. Shen, and M. Kreitman. 1995. A population genetic test of
selection at the molecular level. Science 270:1497–1499.
Thompson, J. N., and B. M. Cunningham. 2002. Geographic structure and
dynamics of coevolutionary selection. Nature 417:735–738.
Tian, D. C., H. Araki, E. A. Stahl, J. Bergelson, and M. Kreitman. 2002.
Signature of balancing selection in Arabidopsis. Proc. Natl. Acad. Sci.
USA 99:11525–11530.
Tiffin, P. 2004. Comparative evolutionary histories of chitinase genes in the
genus Zea and the family Poaceae. Genetics 167:1331–1340.
Tiffin, P., and B. S. Gaut. 2001. Molecular evolution of the wound-induced
serine protease inhibitor wip1 in Zea and related genera. Mol. Biol. Evol.
18:2092–2101.
Tiffin, P., and D. A. Moeller. 2006. Molecular evolution of plant immune
system genes. Trends Genet. 22:662–670.
Tiffin, P., R. Hacker, and B. S. Gaut. 2004. Population genetic evidence for
rapid changes in intraspecific diversity and allelic cycling of a specialist
defense gene in Zea. Genetics 168.
Tishkoff, S. A., R. Varkonyi, N. Cahinhinan, S. Abbes, G. Argyropoulos,
G. Destro-Bisol, A. Drousiotou, et al. 2001. Haplotype diversity and
linkage disequilibrium at human G6PD: recent origin of alleles that
confer malarial resistance. Science 293:455–462.
Wilkes, H. G. 1967. Teosinte: The closest relative of maize. Bussey Institute,
Harvard Univ., Cambridge, MA.
Wright, S. I., and B. Charlesworth. 2004. The HKA test revisited: a maximumlikelihood-ratio test of the standard neutral model. Genetics 168:1071–
1076.
Zangerl, A. R., and M. R. Berenbaum. 2003. Phenotype matching in wild
parsnip and parsnip webworms: causes and consequences. Evolution
57:806–815.
Zhang, L., A. S. Peek, D. Dunams, and B. S. Gaut. 2002. Population genetics
of duplicated disease-defense genes, hm1 and hm2, in maize (Zea mays
ssp. mays L.) and its wild ancestor (Zea mays ssp. parviglumis). Genetics
162:851–860.
Associate Editor: H. Hoekstra
EVOLUTION DECEMBER 2008
3081